IEC62059标准在智能电能表可靠性预计与考核验证方法上的应用_英文_

IEC62059标准在智能电能可靠性预计与考核验证方法上的应用_英文_ IEC62059标准在智能电能表可靠性预计与考核验证方法上的 应用_英文_ Research & Exploration IEC 62059 Standards’ Application in Reliability Prediction and Verification of Smart Meters By Li Xiangfeng and Zong Jianhua Abstract: This article introduces the current situation of the smart meter’s reliability and the failure mechanisms at ? rst, and then describes the relationship of meter reliability characteristics combined with its Bathtub Curve. It also introduces both the feasible failure tree model for meter lifecycle prediction based on actual experiences and meter reliability prediction methodology by SN 29500 norms based on this model. This article also brings forward that it is necessary that the “Learning Factor” shall be adopted in meter reliability prediction for new materials, new process, and customized parts by referring to GJB/Z299C. Thereafter, this article also tries to apply IEC 62059 and JB/T 50070 to introduce the feasible method for the lifecycle prediction result veri? cation by accelerated lifecycle test. Furthermore, the article also explores ways to increase the ? rmware reliability in smart meter. Key Words: Smart meter, reliability prediction, accelerated lifetime test, truncate sequential trial method for reliability test 1. Introduction Since the U.S. made Energy Independence and Security Act in 2007, major economies in the world successively put forward their plan for smart grid to cope with the increasingly serious energy crisis. The STATE GRID Corporation of China (SGCC), based on its own grid characteristics , in accordance with the overall goal of “Security, economy, energy-saving, scientific supervision & control and value-added services ", has vigorously promoted the “ Robust Smart Grid " construction. From December, 2009 to March, 2012, SGCC have bought 150 millions of smart meters with 10 rounds of centralized bidding, and 85% of meters have been installed in field. Although both SGCC and suppliers paid special attention to the quality and reliability, more than 5% of installed meters have to update their firmware or be replaced with new meters, especially in the period of 2010~2011. Therefore, we must not only pay attention to acceptance tests and quality control in all processes such as design, production process, procurement, logistics, installation and shift management, but also enhance the reliability indicators assessment. It is essential for China electrical meters industry to understand IEC 62059 series standards combined with GJB/Z299C and other international reliability standards (such as Mil_HDBK_217F / Mil_STD_883F/TR-332 / SN29500 / IEEE1413 / RDF2000, etc.) to establish an evaluation standard suitable for our own industry. 2. Smart meter lifetime parameters To evaluate the lifetime of smart meters, we should start from the speci? c indicators in respect of reliability engineering. Reference to IEC 61709 and IEC 62059 standards are expressed as follows: 2.1 Failure Failure means the termination of the ability of a meter to perform a required function—the meter is unable to perform 76 CHINA STANDARDIZATION May/June 2012 Research Exploration its required function and unrepairable. In order to classify the malfunction level of smart meters, the faults are classi? ed into A / B / C: A-class fault: measuring stop or measuring inaccuracy, LCD does not display, energy register suffers "digits distortion" or “digits holding", RTC disorders, internal ? re, etc. B-class fault: The display of the LCD is not complete, LCD blinded, meter enclosure broken, billing data or event log missing, internal relay stuck, communication disconnected. C-class fault: color fade on nameplate or meter box, sealing performance declines, metal parts damaged, indicator light or pulse output malfunction, low voltage in battery, the replaceable communication module damaged. If the small defect will not affect the normal usage, it can be neglected from failure, for example: LCD backlight or LED indicators do not shine. 2.2 Failure Rate λ Failure rate is the accumulated times that the unrepairable failure occurs within a set time interval. It is the function of the time, while its unit is Fit (Failure in time: 10-9Failures/h) λ. In IEC 62059-31, failure rate λ(t) is de? ned as instantaneous value: meters can be divided into three periods: Infant Mortality, Useful Life and the Wear-out. The failure rates versus time of each period follow the bathtub curve, shown in Figure 1. We can see MTTF only represents the time of useful life. 2.5 Reliability Reliability R(t) is the probability that meter can perform a required function under given conditions for a given time interval, and it is changing as the time goes. The meter failure is random in useful life period, and the reliability indicators will accord with exponential distributions: Figure 1: Failure rate of electronic equipments and operation time (Bathtub Curve) Infant Mortality: Due to poor quality control in manufacturing process, the early failure rate appears to be ‘high at ? rst, getting lower then’, which is closely related with the quality management level of the enterprise. The higher the level of enterprise management is, the shorter the process will be. Most of the early failure can be eliminated in factory by aging (Burn-in) or environmental stress screening (ESS). But if the strict process control and 6δ management are widely adopted in all the manufacturing process, this problem will be greatly eliminated.Useful Life: The failure of smart meters during this period is random and relates to their intrinsic quality level and environment. The better the quality is, the lower the failure rate will be. The failure rate is very low, so its service life is quite long. Wear-out: The failure intensity becomes unacceptable , all meters shall be replaced /shifted due to the major components lifetime wear out. Various faults will cumulatively break out.2.4 MTTF—Mean time to failureIn useful life period, the failure rate is approximate to a constant. Generally, the reciprocal of the failure rate λ, or mean time to failure (MTTF) to judge the quality of a meter: MTTF is often used to describe the meter lifetime in China. In fact it is inappropriate. It is seen from the above formula that the reliability remains only 36.8% when the meters operation time reach MTTF. In another words, if 100 meters work to the end of MTTF, less than 37 meters will work well. Considering the LCC, the meters operation time when reliability is down to 97% is defined as the meters useful lifetime by British OFGEM. If it is expected that the useful lifetime reach to 10 years, the annual failure rate should not be greater than 0.3%, MTTF=328 years! 90% reliability is enough for 3-phase smart meters and smart terminals which are compulsorily required to periodical veri? cation. In it F(t) and f(t) are respectively the distribution function and the probability density of the instantaneous failure, and R(t) is the reliability function, related to the reliability. An estimated value of the instantaneous failure rate can be obtained from the actual use statistics (see IEC 62059-21), or refer to the simulation test data from authoritative organizations.2.3 Lifecycle & Bathtub curve The reliability during the lifetime of 3. Reliability prediction of IEC 62059-41 smart meter 3.1 The reliability prediction method There are many international and Chinese standards focusing on the electronic equipments reliability prediction. Some well-known standards are: GJB/Z299C, Mil_HDBK_217F, RDF 2000 and IEC61709, IEEE1413, SN29500, Bellcore SR-332, etc. The failure rates and the environmental stress CHINA STANDARDIZATION May/June 2012 77 Research & Exploration of electronic components and processing craft under reference conditions are all provided by these standards. The data of military reliability standards are more conservative and not suitable for the meter reliability prediction. IEC 62059, IEC 61709, LandisGyr ALEG,OFGEM are all based on SIEMENS SN 29500 standards since SN 29500 was updated frequently, covering a lot of fields of industry, medical care, communications, etc. It is recommended to do the failure mode and effect analysis(FMEA)and failure tree analysis (FTA) first in the design stage to estimate the reliability of the meters. In OFGEM approval test, the “worst case” prediction is adopted, all components are simpli? ed to serial failure mode and assessment is made. components in different stages can be found in SN 29500 : learning factor with reference to GJB/Z299C for some components. Learning Factor π L: — component failure rate under reference conditions — the components voltage ,current , temperature dependence factors — the component quality factor —sum of the materials and components 3.2 Learning Factor π L Considering the status of metering industry in China, if the meters to be assessed using the “alternative” or “compatible” materials and components of the brand to do the prediction according to SN 29500, it is need to introduce the 3.3 Lifetime prediction of dedicated components Some dedicated parts , customized by meter manufacturers, such as enclosure, metal parts, LCD, the lithium-ion battery, CT sensor, transformers, relays, or breakers, etc, are not included in common reliability books. If the failure rates and stresses of dedicated components cannot be found in SN 29500 or SR-332, it is better to ask the components suppliers to provide the 3rd party quali? cation data after authoritative testing. The learning factors of them shall refer to Table 1. Figure 2 : OFGEM simple series modeAccording to the OFGEM simplified model, the total failure rate of the meter λis equal to the sum of the failure rate of all components and process methods and the stress factors. The components failure rate λ_ref and its corresponding stress factors π, the failure rate of all Table 1 : Learning factor πL 78 CHINA STANDARDIZATION May/June 2012 Research Exploration As for critical parts, such as Li-ion battery and LCD which are not included in existing norms (it doesn’t mean the components suppliers have no their own statistical data) but determine whether meters can work with the expected life, the manufacturers are required by OFGEM that no quality problem will come into being beyond the certified lifetime. 3.4 Lifetime certi? cationLife time certi? cation in China is still in exploratory stage now, China EPRI or other authorities can certify as below:a. Random sampling smart meters from meters factory by GB/T 2828.1-2003 or ISO 2859-1:1999. b. Maximum 10-year lifetime certificated while the sample meters pass all function test and type test. The certification is done according to IEC 62059-41. c. The smart meters suppliers who expect to get more than 15-year lifetime (97% reliability) certificates after 1 year mass-production, and the reliability prediction data by IEC 62059-41 is more than 15-year (reliability 97%), the authorities (CEPRI) shall do accelerated life test according to IEC 62059-31. d. The highest certificated life is 20 year. d. Confirm compliance with the reliability indicators of smart meter speci? cations. There are many standards can be used to do the reliability test of smart meters as electronic measuring equipment, such as: I. China machinery industry standards represented by JB/T 6214 and JB/T 50070 standards. II.IEC 62059-31 accelerated lifetime test method. ?. LANDISandGYR ALEG evaluation method. IV. Relatively strict US military standard Mil-STD-883 and Mil-STD-810.4.1 JB/T50070 assessment methodsJB/T 50070 standard is based on GB 5085.5 (IEC 605-5), JB/T 6214-91 and ambient temperature cycling simulation (ESS) test. It is effective to those meters whose accuracy will change due to ambient temperature varied and pulsed load. JB/T 50070 assessment method generally can cover 80% of failure types of electromechanical meters and about 50% of static meters. Besides accuracy issues, the meters, especially the electronic electricity meter, fault modes and causes are various. For instance, semiconductor components failure caused by sudden high-energy electromagnetic radiation, LCD wear-out due to UV and cosmic radiation, E2PROM writable cycles run-out, terminals corrosion, and so on. But all above factors are not considered in JB/T 50070 yet. 4.2 IEC 62059-31-1Accelerated Lifetime Testing (ALT) Accelerated Lifetime Testing is recognized by most of international reliability standards. It can realize the testing of the meter reliability and lifetime in much shorter time. The differences compared to JB/T 50070 are: i. Based on the molecular kinetic theory that most major failures are result of the molecules activation energy activated by environmental factors which changes the original stable molecular arrangement. ii. Temperatures increasing can raise the molecular activation energy, thereby force meters to enter "accelerated" failure state. iii. Accelerated testing can signi? cantly reduce the base of testing samples and shorten test duration; besides, it is more economic as well. The Arrhenius temperature acceleration model can be used for indoor electricity meters, as the indoor temperature fluctuates slightly. So the Arrhenius temperature acceleration model is recommended: 4. Reliability test by IEC 62059-31 Reliability test is very useful for meters acceptance and meters reliability improvement. The benefits are summarized as follows: a. Find out the defects and weakness in design, components quality and manufacturing process, and provide guidance for reliability improvement. b. Provide information to improve the performance, reduce maintenance cost and accessories consumption, or provide basic reliable data for new components which were used in design. acceleration factors (AF): tu—time interval to failure of a meter at normal use temperature. ts—accumulated testing time to failure of a meter at stress temperature. E A—activation energy (0.3~1.5eV),generally 0.9eV: K—Boltzmann constant ( 8.617 x 10-5 eV/K ) : Tu—Average temperature at normal use. unit K Ts—Average temperature at stress temperature. unit K The Hallberg-Peck (Eyring) accelerated model can be used for the moisture environment: RH u—the percent relative humidity at normal use; RH s—the percent relative humidity at stress conditions; Generally n=3, E A=0.9eV; Special attention must be paid in accelerated test process: a. Investigation on all factors possibly causing over stress and failure should CHINA STANDARDIZATION May/June 2012 79 Research & Exploration be conducted before accelerated testing. Select the test items which are most likely to lead to failure with time as the accelerated parameters. Meter accuracy is most likely to change over time; therefore the focus should be on accuracy, and verify the other parameters (RTC accuracy, voltage, current, functions, etc) by the way. b. It doesn’t mean the bigger the AF, the better it is. The stress level defining (temperature or/and humidity) must avoid direct damage by over-stress. c. To increase the temperature and humidity will lead to meter terminals and metal parts corrosion increased, also cause the enclosure performance decline. However, since the accuracy is the main acceleration index, if the corrosion has not caused the main function fault, the ALT can be continued. For a certain type of smart meter, for example, its average annual operating temperature is 25 ? C, average humidity of 50%, operating temperature limit of -30 ~ 85 ? C, using temperature and humidity stress for accelerated testing, respectively, verify 10 years, 15 years and 20 years of available time, cumulative time of the accelerated test can be found Table.2: Peck Model example (Tu=298K,RHu=50%) Increasing the quantity of meter samples will greatly reduce the accelerated life test duration. The random sampling methods is defined in GB/T2829, and the eligible determinate comply with truncate sequential trail method defined in JB/T 6214. Table.3 is 4:6 Eligibility determinant table for truncate sequential trial (α=β=0.2, Dm=2)If the test equipment allows, we can apply both the variable voltage (0.7Un~1.2Un) and 100s-ON 20 s-off Table 3 : 4:6 Eligibility determinant table for truncate sequential trial During ALT, voltage Un and current are applied: A. Direct connection meter: I=0.1Imax or Ib B. CT connection meter: I=0.5Imax or I n 80 CHINA STANDARDIZATION May/June 2012 Research Exploration pulsed load(0~0.5Imax ) during the ALT process. In addition, if the accelerated life test is in conjunction with Weibull graph papers, we can furthermore to gain the full life cycle curve of the test samples. It can provide guidance to determine whether the same batch meters shall be used continuously or be shifted right now. High acceleration test (HAST) has been the hot topic for recent years. According to SSB-1004, we can do destructive continuous THB effect accelerated test at 130?,85,RH. It is useful to evaluate the overall quality indicators of semiconductors in a shorter time. But we think it is not fit for smart meters as the later have frail crusts. 5. Software reliability by IEC 62059-51 Besides of the hardware, another key factor affecting the reliability of the smart meter is the reliability of embedded software. About 55% to 70% of failures of smart products are caused by software/ ? rmware according to statistics. All smart meters rely on much complex ? rmware to complete basic metering and prepayment function expect mechanical and simple electronic meters, so the reliability of software is very critical .The more functions smart meters have, the more complex the firmware is, and the higher failure rate will happen. Software reliability is the ability that the software system can realize the specified functions within the specified time and under environmental conditions. The smart meter software reliability is not only related to its bugs, but also to the smart meter specifications(norms) perfection, design input requirements ? le exactness, configure files correctness, AMR system reliability, users pro? ciency, etc. A reliable software system should be correct, complete, consistent and robust. However, no software is 100% perfect in fact, and its correctness is not easy to measure accurately. Therefore, from a software engineering point of view, we can only measure its reliability through the software debugging ("white box" and "black box" debugging combined). As the development period of software is relatively long, the maturity of software, to some extent, determines the maturity of a product. The validation test of protocol DL/T 645-2007 operated by China EPRI recently is a useful exploration of meter software reliability by“Black box” debugging. 6. Conclusion Although the exploration in smart meters reliability started late in China, the quality and international competitiveness of China’s electrical meter industry will be improved as the implementation of GB/T 17215.9xx-2011 will certainly push the upgrading of the industry as a whole. About the Authors CHINA STANDARDIZATION May/June 2012 81